2020/05/04 by Souvik Chakraborty, Chakraborty, Souvik · 1 citation
Decision Sciences · Engineering · Physics and Astronomy · #FOS: Computer and information sciences #Machine Learning (stat.ML) #Model Reduction and Neural Networks #Nuclear Engineering Thermal-Hydraulics #Probabilistic and Robust Engineering Design
paper · pdf · doi:10.48550/arxiv.2005.01302
openalex publication_date 2020/05/04 · openalex created_date 2020/05/13 · openalex updated_date 2026/07/28
This paper presents a simulation free framework for solving reliability analysis problems. The method proposed is rooted in a recently developed deep learning approach, referred to as the physics-informed neural network. The primary idea is to learn the neural network parameters directly from the physics of the problem. With this, the need for running simulation and generating data is completely eliminated. Additionally, the proposed approach also satisfies physical laws such as invariance properties and conservation laws associated with the problem. The proposed approach is used for solving three benchmark reliability analysis problems. Results obtained illustrates that the proposed approach is highly accurate. Moreover, the primary bottleneck of solving reliability analysis problems, i.e., running expensive simulations to generate data, is eliminated with this method.